The Spirit Level’

Provided by the author(s) and University College Dublin Library in accordance with publisher policies. Please cite the published version when available.

Title Affluence versus Equality? A critique of Wilkinson and Pickett’s book ‘The Spirit Level’

Authors(s) O'Connell, Michael F.

Publication date 2010-09

Item record/more information http://hdl.handle.net/10197/2475

Downloaded 2021-09-26T14:03:07Z

The UCD community has made this article openly available. Please share how this access benefits you. Your story matters! (@ucd_oa)

Working paper, September 2010

Dr. Michael O’Connell ([email protected]), Senior Lecturer, School of Psychology, UCD1, Dublin 4, Ireland

Summary The Spirit Level made strong claims that in developed countries, income growth was no longer important and the focus should turn to income differentials within society. Putting affluence before parity and solidarity led to the rise of widespread anxiety, insecurity and social dysfunction. In this paper, six problems are identified with the argument made in the Spirit Level: 1. There is no conflict between wealth and equality. In fact they tend to be highly correlated (i.e. wealthy societies are far more egalitarian than poorer societies); 2. Correlational data are relied upon to imply causal direction; 3. The focus on income inequality ignores the role of savings and state services; 4. There is no evidence that people are systematically stigmatised by buying ‘second-class’ goods; 5. Investment in ‘luxuries’ in wealthy countries have unforeseen consequences in raising living standards in poorer countries, e.g. the mobile phone; 6. The status of women is far more highly associated with a country’s wealth than its level of inequality.

1 The views expressed in this paper are those of the author, and do not necessarily reflect the views or policies of UCD or the School of Psychology.

1 The capitalist achievement does not typically consist in providing more silk stockings for queens but in bringing them within the reach of factory girls in return for steadily decreasing amounts of effort … The capitalist process progressively raises the standard of life of the masses: Joseph Schumpeter, Capitalism, Socialism and Democracy, 1942, p. 67.

As countries make the transition from low to middle income, most people gradually no longer have to worry about the availability of food, there are improvements in hygiene, and longevity, and health standards increases rapidly. Satisfaction with life is found to grow substantially in these times of rapid change, and education levels for boys and then girls are raised. And with greater literacy, virtuous circles of growth and knowledge, and perhaps even democracy kick in. As Wolf notes, “social progress [including the liquidation of child labour] has been greatest where incomes have risen fastest” (2004).

The Equality Case against Affluence But what if there are limits to the connections between economic growth and positive social outcomes? Is there a point at which we just have enough ‘stuff’, and further abundance is wasteful and counter-productive? This is precisely the argument made by Richard Wilkinson and Kate Pickett in their 2009 book, entitled ‘The Spirit Level’. The book was extremely well received – “a compass to rebuild our society” (Johann Hari), “brave and imaginative” (Michael Sargent), “deserves the widest possible readership” (Iain Ferguson), “the most important book of the year” (John Crace), and “profoundly important” (Lord Layard). The authors are academics and gather data to resolve a conundrum - despite living at a time of unprecedented human productivity, people are trying to cope with great levels of stress, alienation and unhappiness. Their explanation is that overall income growth is no longer important in developed countries, rather the focus should turn to “income differences within … society” (p. 11). Because of their short-sighted focus on growth over equality/inequality, modern societies have become increasingly dysfunctional. Whereas when countries start to get richer, economic growth is hugely important in raising living standards and combating the diseases of poverty like tuberculosis, cholera and measles, once a

2 good standard of living is attained for the majority, then “countries inevitably reach a level of affluence where ‘diminishing returns’ set in” (p. 10) – further growth contributes less and less, and the diseases of affluence spread. ‘The Spirit Level’ has an explicitly anti-growth message – “we have got close to the end of what economic growth can do for us … economic growth for so long the great engine of progress, has, in the rich countries, largely finished its work” (p. 5). The focus of policy makers now needs to be on the growth of inequality, the cause of “the broken society and the broken economy” (p. 5). “The problems in rich countries are not caused by the society not being rich enough (or even by being too rich) but by the scale of material differences between people within each society being too big” (p. 25).

The obsessive pursuit of growth and affluence by policy-makers is matched by the problems of consumerism and ‘affluenza’ among the general public – “consumerism … makes it so much harder to contain economic activity within sustainable levels. Our addiction to shopping and spending makes many people think we have already lost the battle against global warming” (p. 221). Our shopping takes on huge social-psychological significance it is argued, since “second-rate goods are assumed to reflect second-rate people” (p. 30).

As noted above, this book made quite an impact. Unlike say the poorly substantiated muddle that is Oliver James’ Affluenza (2007) – a Sunday supplement article puffed into a book, as one reviewer noted - this is a significant work by serious authors with, especially in the case of Wilkinson, a considerable track record of publication and reflection in the area. The authors are absolutely sincere about their approach as a scientific one based on solid data. In fact they tell us in the preface (page ix) that they were originally going to call the book ‘Evidence-based Politics’, (to which we can only respond, thank God for editors). So we can consider ‘The Spirit Level’ as an expert witness called to testify against affluence. And their charge sheet makes for some grim reading. In the race for untrammelled growth in many societies, equality has been neglected. The consequence of putting affluence before parity and solidarity has led to the rise of widespread anxiety, insecurity, evaluation anxiety and self-promotion, while

3 the positives in society have been in decline – community, self-deprecation, modesty and secure social selves. Trust and social capital have declined and women’s social status relative to men’s are under attack. In commodity-rich but precarious contemporary societies, people’s stress levels affect both their mental health levels but also their physical health. Furthermore the inequalities associated with modernity lead people to adopt unhealthy lifestyles that promote obesity and alcoholism, thus further undermining their health. Educational performances fall, teenage pregnancies and family dysfunctions rise, youth crime, public disorder and violence all grow.

The authors include a cartoon (p. 214) where the key problem is depicted; by Goddard, the illustration shows three young women speeding in different directions along the High Street in a frenzy of shopping. All three carry huge shopping bags with unsubtle labels like ‘Really ostentatious big bag’ and ‘I obviously shop somewhere really expensive’ and are wearing sunglasses pushed up over their hair. One of the women is shouting down her mobile phone. We are meant to infer that her phone conversation - like her shopping - is entirely frivolous (she’s not ordering a copy of ‘The Spirit Level’ over the phone for instance). This, for the authors is the problem in a nutshell – the returns to society for these ever more flashy consumerist and materialist lifestyles are not widespread satisfaction and solidarity, but all the problems and dysfunctionality noted above. The battle to achieve gender equality, to raise educational standards, to tackle mental and physical health challenges, to create communities that really work, are all undermined, not helped, by economic affluence. And you can add to that the binge-drinking excesses and street violence in so many towns across Europe every weekend. A one-dimensional addiction to the acquisition of goods is destroying our planet’s ecology. You can see why Wilkinson and Pickett are happy to conclude that “almost every modern social and environmental problem – ill-health, lack of community life, violence, drugs, obesity, mental illness, long working hours, big prison populations – is more likely to occur in a less equal society … [we need] to shift the balance from self-interested ‘consumerism’ to a friendlier and more collaborative society” (from blurb on hardback book cover).

The ‘Spirit Level’ Thesis – Some Quibbles But hold on, we need to look more closely at how these conclusions are reached. The first thing is just how weak the evidence provided by Wilkinson and Pickett is. They claim to use their “training in epidemiology” (p. ix) and the methods employed in that discipline to ensure the readers are provided with the best scientific evidence, with observable and objective outcomes, and so on – putting the evidence back in to ‘Evidence-based Politics’. (This is their analogy to ‘Evidence-based medicine’.) The results will not be “some statistical trick done with smoke and mirrors” (p. 23). But in fact they largely confine themselves to using the very simplest method, that of correlation. This is of course a very widely used technique – it is also a very basic one - for example, correlation has just been introduced into the maths syllabus for fifteen year olds in Ireland. Nothing wrong with simplicity of course, except that because correlation is restricted to a measurement of association between two set of scores, it can tell us absolutely nothing about causation. In other words, if A is correlated with B, we have no way of knowing if A causes B, if B causes A, if a third measure C causes both A and B, or it’s a coincidence. In most introductory statistics courses, the first lecture is given over to teaching people how to do correlations, and the next four to explaining why ‘correlation is not causation’. Studies in human reasoning show we have a tendency to make causal attribution very (too) readily when things seem to go together. An example of two things that correlate together but are not causally linked is helpfully provided by Wikipedia. If you search for the Wiki page called (yes, you guessed it), ‘Correlation does not imply causation’, at http://en.wikipedia.org/wiki/Correlation_does_not_imply_causation there is a diagrammatic presentation of the link between tons of fresh lemons imported from Mexico to the US per annum, and the US Highway Fatality rate. The relationship is almost perfect (r = - 0.99), as lemons increase, fatalities decline. This is of course perfectly spurious, or rather coincidental. We should not conclude that citrusy freshness saves lives. Nor, in springtime, as the number of birds increase, and the grass grows greener, should we assume a causal

5 relationship – rather it’s the increasing amount of sunlight that influences these independent outcomes.

But of course ‘everyone knows’ about the correlation and causation issue. Well, the problem for Wilkinson and Pickett is that the evidence they provide in The Spirit Level is no stronger than this, i.e. virtually all we are presented with are correlations. This evidence should only be considered the most basic first step in building their causal argument. And even some prominent readers do not seem to get this – here’s a quote from a rave review of the book from Colette Douglas Home writing in the Glasgow Herald (available on the webpage linked to The Spirit Level) – “graph after graph demonstrates cause and dire effect”. Sorry Colette, they don’t. As correlations, logically, they cannot demonstrate cause and effect, dire or otherwise. They are suggestive at best, a point that Wilkinson and Pickett do not acknowledge clearly enough.

You might wonder then what researchers are meant to do – how can they establish causality in relationships they study? Some find experiments useful because they can randomly assign the objects of our study to different conditions. But we can’t randomly assign a society to be very inegalitarian for a decade and another one to be quite egalitarian, and then compare and contrast. Instead very sophisticated statistical techniques –such as multi-level modelling where cross-sectional and temporal (country and time period) measurements are simultaneously assessed tend to be employed. These are routinely used in the top epidemiological journals but make no appearance in The Spirit Level. But perhaps it is unfair to expect the authors to introduce very sophisticated methods to make their case. Are we just being sticklers here? Doesn’t merely eyeballing the graphs in The Spirit Level make it abundantly clear that equality is the key causal force? Let me give one very practical example why not. In chapter eight of The Spirit Level, a correlation is reported between educational performance and social equality. The OECD’s programme of comparative international educational – ‘PISA’ – consistently show that a representative sample of 15 year olds in a very egalitarian society like Finland do better in tests than a sample from an equally wealthy but less egalitarian society. But is this

6 because Finland is a more egalitarian society? An alternative answer is offered in an article written by Professor Alan Smithers, a UK expert on comparative educational performance, “Finland heads the published rankings because it has very few new arrivals, i.e. low immigration. But if only native children (those born in the country of assessment to at least one parent born in that country) are considered, the Netherlands obtains the top score in maths”. (Guardian, 14th December, 2004). In other words, because children of immigrants often face additional challenges in the educational system in terms of poverty, language skills and their parents’ lack of knowledge of the way the system works, their presence in higher numbers translates into a lower country performance. Immigrants are disproportionally poorer than the native population so their presence also increases (temporarily) levels of income inequality. The dip in educational performance and the higher level of inequality of the Netherlands are both influenced by a positive third factor – openness to immigration. But one gets no sense of this from the Spirit Level’s cursory analysis and correlational approach.

There are other problems with The Spirit Level thesis. The focus throughout the book is on ‘income inequality’, that is on the parity of cash income available to households. But there’s a lot more to understanding economic resource differences than income. The OECD published a landmark study of inequality in 2008 - called ‘Growing Unequal?’ – which interpreted data from across all its member countries in conjunction with leading economic scholars and national consultants. In ‘Growing Unequal?’ the question is asked, “Can we assess economic inequalities just by looking at cash income?”(p. 294). And can we? “The short answer to this question is ‘no’” (p. 294). People can enjoy quite high living standards even if their incomes are low – governments may provide key services like education and health free at the point of demand. In fact in recognition of this, the EU gathers two kinds of data on equality – one before and one after ‘social transfers’ from government. The difference between these measures is substantial. As ‘Growing Unequal?’ puts it, “the effect of government services in narrowing the Gini coefficient of income inequality is quite large” (p. 295), and similarly, “measures of inequality based on annual cash incomes

7 overstate the overall inequality of resources available to households” (p. 301). Savings also matter. People with low incomes may have high living standards because they own assets (such as a house with the mortgage paid off). “The older people are, the less likely they are to be deprived materially and the more likely they are to hold greater assets”(p. 301). Furthermore, income measures at point can be very static. If researchers follow the lives of individuals over time in a more dynamic framework, the stereotype that the same people are poor all of the time is challenged. “Around 45% of people counted as poor based on static income will move above the poverty threshold over the next years”. (p. 299). In other words, there are few nice things to say about poverty but one good thing is that a lot of people who are poor at any point don’t stay poor for long.

Beyond Quibbles - The 800 Pound Gorilla in the Room So there are some problems with the Wilkinson and Pickett thesis – they imply causal relationships without strong enough evidence, they use income as their measure of resources and they work with a rather static concept of poverty. But really these are quibbles, and their work certainly is suggestive of a conflict between affluence and equality. Until one stops and thinks about the following – if a book is published making a claim that economic growth has done its work, and that further growth is a bad thing as it undermines equality of income, and if close to sixty correlations are used in that book to make the argument, then is it not suspicious that the correlation between growth and equality is not explored in detail? Because guess what – and this is the kicker – if you were to adopt the same approach as ‘The Spirit Level’ and analyse the co-relationship of economic growth and equality, it turns out they actually go pretty well together. You can show that affluence actually creates equality … whoops, have to be careful with that causality claim, I mean you can show they correlate very well.

Let’s estimate the correlation - I use precisely the same 23 countries as those selected by Wilkinson and Pickett and listed in their appendix (The Spirit Level, p. 267), and select the same measures, i.e. those “provided ready-made by the United Nations” (The Spirit Level, p. 18) including “the Gini co-efficient: it is the most common measure [and] it is favoured by economists” (the Spirit Level, p.

8 18). The data for the 23 countries are taken directly from the United Nations Human Development Report (2007/08): for GDP per capita, column three, page 277 (table 14, ‘Economic performance’) and for inequality, the Gini index, column eight , page 281 (table 15, ‘Inequality in income or expenditure’). The correlation (Pearson’s r) comes out at -0.526. This means that growth in GDP per capita is associated with a decline in inequality, or putting it more simply, more affluence is associated with more equality. I probably should repeat that important finding: more affluence is associated with more equality. The probability of a result like this happening by chance is exactly 1 in a 100 (p=0.01) which comfortably meets the statistical norm of being a ‘significant’ relationship. So what’s all this brouhaha about growth and its relationship to social equality? They are complementary, not in conflict – countries that are wealthier are on average more equal. But you would never know anything about this relationship from reading The Spirit Level. Perhaps it really is just as well that Wilkinson and Pickett didn’t call their book ‘Evidence-based Politics’. ‘

Furthermore, this relationship between growth and equality is – alas – not my discovery. The relationship is so well-known by economists that it has its own name, the Kuznets curve, after the Russian-American economist Simon Kuznets. It shows that fairly predictably in a country undergoing development, income inequality typically widens at first, levels off and then declines. Economists have found that the Kuznets model makes quite accurate predictions about the relationship between growth and equality, see for example Robert Barro’s article (2000) in the Journal of Economic Growth or the more recent paper by Thorsten Beck and colleagues (2005). Despite the prominence and status of the curve positing good news about the link between growth and equality, one searches in vain for any reference to it in The Spirit Level.

Recent Trends in Inequality There is one valid concern about socio-economic trends made in The Spirit Level and this is that overall there has been a move towards greater inequality in the last two to three decades in the affluent countries. That is uncontested by any observer. The question is why this has occurred. Ben Friedman (2005) has

9 argued convincingly that changes in income inequality in the advanced countries are largely due to an interplay between technological development and workers’ human capital (or skills). The shift from industrial production in the late 1970s, and the rise of new technologies such as computing put a premium on IT skills. A skill imbalance was generated whereby information-based services accelerated opening a wide earnings gap between ‘knowledge’ workers who had this expertise and traditional workers who tended not to. In time, and with the spread of education and the wage and employment incentives associated with computing, that imbalance was righted (or even overshot, as a glut of IT trained workers were produced while shortages in skilled trades-people brought the much-needed Polish plumber to Western Europe in significant numbers). Of course, where gaps do open up in earnings for periods of time, the high earners in virtually every developed country contribute heavily in their taxes to the public purse, a good deal of which goes to support public welfare and public education.

However, the trends towards inequality in recent years should not be overstated. The Spirit Level suggests that countries like Sweden “have had steep rises [in inequality] since the early 1990s”. (p. 236). The OECD’s ‘Growing Unequal?’ on the other hand includes Sweden among countries that have had a small increase in inequality from the mid 1990s to the mid-2000s (table on p. 286). Overall, the OECD study is far more measured in its account of change – and argues that the increase in inequality “has not been as spectacular as most people probably think is has been … [the difference is] not one that would justify talking about the breakdown of society” (p. 15). It estimates that in terms of income equality in 27 OECD countries, a comparison between the mid 1980s and the mid 2000s, a slim majority (14) have registered no change, eight have shown a ‘small increase’ in inequality, two a ‘significant increase’, and three a ‘small decrease’ (table 11.1, column A3). Given how authoritative the OECD study is of inequality, and that its publication date was 2008, it is surprising that The Spirit Level published in 2009, made no reference to it. But then given how sober and balanced an account of inequality ‘Growing Unequal?’ offers, perhaps it is not that surprising.

10 And the basis for movement towards lesser or greater social inequality is quite subtle. Technological and labour market shifts, as noted above, play an important role. So too, argues the OECD study, is changing demography. There has been a trend in the developed countries of ever smaller households, meaning that “economies of scale in consumption are lost and that a higher money income is needed to assure the same level of household well-being” (2008; 288). Increased longevity (surely a good thing?) can mean more elderly people living as lone households, while the widespread trend towards single parenting also contributes; “these structural factors have increased income inequality in a majority of countries and significantly so” (2008; 288).

Second-class goods? Even a basic motor of change proposed by the enemies of affluence makes no sense if we think carefully about it. Part of the implicit Spirit Level thesis is an obsession with consumerism – the latter authors pinpoint “our consumerism [and] addiction to shopping and spending” (p. 221). One has to wonder whether these people have tried high-street shopping recently? Am I alone in thinking it borders on the hellish? Shopping, like air travel, perhaps once conferred a kind of jetset mystique, a bit like cheesy-pineapples on a stick. But now the negotiation of a busy shopping mall, or a thronged airport is about as glamorous as, well, cheesy-pineapple appetizers. Most of us, I contend, dislike shopping, but do it because we have to, because the kids demand Cheerios, the baby needs Pampers and the cat gets pretty mean without her Whiskas. Of course, there is also a small but determined subset of people who still like air travel or shopping. But these people should be seen for what they are - strange eccentrics, not a substantial social trend. How do we know this? Because shops have to spend a fortune advertising just to try to get us through their door, and change their look and design every year to persuade us to stay. Our ‘addiction’ to brands is actually so slight that as soon as ‘hard discounters’ like Lidl and Aldi expanded beyond their German base into the European grocery market, they enjoyed spectacular success. ‘Hard discounters’ offer a limited product range and a predominance of low-price generic brands. “Second-class goods make us look like second-class people” argues The Spirit Level, (p. 222) – this is why we’re supposed to be

11 obsessed with luxury goods. Not a bit of it – decent low-priced goods make us look like smart and astute people. Bye-bye Whiskas, hallo Katzen-Imbiss! And of course paying a little more for a brand-name product does not necessarily mean succumbing to ‘Affluenza’. Especially with an expensive product, a brand name means the company is more likely to still be around in a few years time, that the consumer can track down someone to complain to if problems arise, and that there should be an expectation of above-average quality associated with the product.

Does Affluence Raise the Status of Women? More affluent societies are on average more equal societies, and vice versa. This is true taking account of all nations, but it is also true when we limit our considerations only to those developed (i.e. wealthy) ones. This means that all the positive aspects – a more enriched community life, better health outcomes and life expectancy, stronger family life, less violence and punitive justice – that are alleged to be linked to equality may well turn out to be functions also of affluence and growth. We have already seen that the apparent link of educational results to equality is spurious and that instead, more dynamic economies attract more immigrants whose initial low incomes and potential language problems often impact on both the economic and educational outcomes of a country.

What about progress on women’s rights? Is women’s status linked to either income equality or to the affluence of a society? My hunch is that the feminist cause should be helped by both. This can be assessed. Again I select the same 23 countries as above, and obtain the universally-respected measure from the UNHDR report (2007/08, table 28, p. 326, column 2), called the GDI (the Gender Development index). This is a measure of the gender disparity assessed in each country. Higher values indicate less gender disparity (i.e. women are closer in status to men when the score is high). When we correlate the GDI with both wealth (the GDP per capita, source given earlier in this chapter) and the measure of income inequality (Gini co-efficient, source given earlier), we find that, yes both equality and affluence are positively correlated with women’s status. But we can go one step further and ask, as social scientists often have to do when

12 two measures are predictive of an outcome, which is the more important one, controlling for (or taking into account) the other. We use a technique called multiple regression, with the GDI as our outcome measure to be assessed. It tells us that together affluence and equality explain about 36% of the variance – that it, the dispersion – of the values of the GDI. But here’s the really important bit – all the explaining is being done by GDP or affluence. The strength of its relationship with women’s status – even while taking income equality into account - is very strong, while the relationship of equality to women’s status – when taking GDP into account – is trivial. The relevant statistic of relative strength is called the t, and for affluence the t = 2.70, while for equality, t = 0.78. In other words, affluence (GDP) is about three and a half times more important – among the developed countries – than equality (Gini) in predicting better status for women in society. Gender equality is favoured by growth, not hindered.

Equality and affluence are allies in making societies better places to live. But equality is the junior partner in this alliance – most of the hard work in ameliorating hardship is done by affluence, that is by the slow and steady accretion of wealth through decades and centuries of human ingenuity and industry. Any other conclusion is based on sentiment, prejudice or ideology, not on the evidence. And the advanced economies have changed but little in terms of equality in the last few decades, and that change is largely due to technological and demographic shifts.

Revolutionary yuppie toys Let’s conclude by returning to Goddard’s cartoon, cited earlier above – remember, the ditsy ladies who lunch, with their unfeasibly vulgar shopping bags, sunglasses and of course, their mobile phones. It’s that last item I’m interested in, the mobile phone. It has become surely the classic symbol of affluence, upon which we lavish and waste far too much money, time and thought, and in whose myriad of functions, we bask in smug glory. (As Marcus Brigstocke quipped, “to the people with iPhones, you just bought one, you didn’t invent it!”) But these toys for the affluent world have also triggered enormous benefits for the developing world. The mobile phone is revolutionary “in places

13 with bad roads, unreliable postal services, few trains and parlous landlines [and] allow quicker and easier access to information on prices, enable traders to reach wider markets, boost entrepreneurship and generally make it easier to do business” (Economist, September 26th, 2009, p. 1, report on telecoms in emerging markets). Jeffrey Sachs, the guru of development, author of ‘The End of Poverty’ and economics professor at Columbia University’s Earth Institute notes how the mobile phone, once the yuppie toy par excellence is now “the single most transformative tool for development”. Affluence and its products can benefit not just the world’s rich, but the bottom billion too. We cannot accept Wilkinson and Pickett’s proposition that “we have got close to the end of what economic growth can do for us” because quite simply affluence has virtually unlimited potential in pushing innovative technology forward.